skip to main content
DOE PAGES title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Performance Comparison of Knowledge-Based Dose Prediction Techniques Based on Limited Patient Data

Abstract

The accuracy of dose prediction is essential for knowledge-based planning and automated planning techniques. We compare the dose prediction accuracy of 3 prediction methods including statistical voxel dose learning, spectral regression, and support vector regression based on limited patient training data. Statistical voxel dose learning, spectral regression, and support vector regression were used to predict the dose of noncoplanar intensity-modulated radiation therapy (4π) and volumetric-modulated arc therapy head and neck, 4π lung, and volumetric-modulated arc therapy prostate plans. Twenty cases of each site were used for k-fold cross-validation, with k = 4. Statistical voxel dose learning bins voxels according to their Euclidean distance to the planning target volume and uses the median to predict the dose of new voxels. Distance to the planning target volume, polynomial combinations of the distance components, planning target volume, and organ at risk volume were used as features for spectral regression and support vector regression. A total of 28 features were included. Principal component analysis was performed on the input features to test the effect of dimension reduction. For the coplanar volumetric-modulated arc therapy plans, separate models were trained for voxels within the same axial slice as planning target volume voxels and voxels outside themore » primary beam. The effect of training separate models for each organ at risk compared to all voxels collectively was also tested. The mean squared error was calculated to evaluate the voxel dose prediction accuracy. Statistical voxel dose learning using separate models for each organ at risk had the lowest root mean squared error for all sites and modalities: 3.91 Gy (head and neck 4π), 3.21 Gy (head and neck volumetric-modulated arc therapy), 2.49 Gy (lung 4π), and 2.35 Gy (prostate volumetric-modulated arc therapy). Compared to using the original features, principal component analysis reduced the 4π prediction error for head and neck spectral regression (-43.9%) and support vector regression (-42.8%) and lung support vector regression (-24.4%) predictions. Principal component analysis was more effective in using all/most of the possible principal components. Separate organ at risk models were more accurate than training on all organ at risk voxels in all cases. Compared with more sophisticated parametric machine learning methods with dimension reduction, statistical voxel dose learning is more robust to patient variability and provides the most accurate dose prediction method.« less

Authors:
ORCiD logo [1];  [1];  [2];  [1];  [1]
  1. Department of Radiation Oncology, University of California, Los Angeles, Los Angeles, CA, USA
  2. Department of Computer Science, University of California, Los Angeles, Los Angeles, CA, USA
Publication Date:
Research Org.:
RadiaSoft, LLC, Boulder, CO (United States); RadiaBeam Technologies, Santa Monica, CA (United States)
Sponsoring Org.:
USDOE Office of Science (SC)
OSTI Identifier:
1481434
Alternate Identifier(s):
OSTI ID: 1612701
Grant/Contract Number:  
SC0017057; SC0017687
Resource Type:
Published Article
Journal Name:
Technology in Cancer Research & Treatment
Additional Journal Information:
Journal Name: Technology in Cancer Research & Treatment Journal Volume: 17; Journal ID: ISSN 1533-0346
Publisher:
SAGE
Country of Publication:
United States
Language:
English
Subject:
59 BASIC BIOLOGICAL SCIENCES; Oncology; radiotherapy; knowledge-based planning; dose prediction; machine learning; automated planning

Citation Formats

Landers, Angelia, Neph, Ryan, Scalzo, Fabien, Ruan, Dan, and Sheng, Ke. Performance Comparison of Knowledge-Based Dose Prediction Techniques Based on Limited Patient Data. United States: N. p., 2018. Web. doi:10.1177/1533033818811150.
Landers, Angelia, Neph, Ryan, Scalzo, Fabien, Ruan, Dan, & Sheng, Ke. Performance Comparison of Knowledge-Based Dose Prediction Techniques Based on Limited Patient Data. United States. doi:https://doi.org/10.1177/1533033818811150
Landers, Angelia, Neph, Ryan, Scalzo, Fabien, Ruan, Dan, and Sheng, Ke. Thu . "Performance Comparison of Knowledge-Based Dose Prediction Techniques Based on Limited Patient Data". United States. doi:https://doi.org/10.1177/1533033818811150.
@article{osti_1481434,
title = {Performance Comparison of Knowledge-Based Dose Prediction Techniques Based on Limited Patient Data},
author = {Landers, Angelia and Neph, Ryan and Scalzo, Fabien and Ruan, Dan and Sheng, Ke},
abstractNote = {The accuracy of dose prediction is essential for knowledge-based planning and automated planning techniques. We compare the dose prediction accuracy of 3 prediction methods including statistical voxel dose learning, spectral regression, and support vector regression based on limited patient training data. Statistical voxel dose learning, spectral regression, and support vector regression were used to predict the dose of noncoplanar intensity-modulated radiation therapy (4π) and volumetric-modulated arc therapy head and neck, 4π lung, and volumetric-modulated arc therapy prostate plans. Twenty cases of each site were used for k-fold cross-validation, with k = 4. Statistical voxel dose learning bins voxels according to their Euclidean distance to the planning target volume and uses the median to predict the dose of new voxels. Distance to the planning target volume, polynomial combinations of the distance components, planning target volume, and organ at risk volume were used as features for spectral regression and support vector regression. A total of 28 features were included. Principal component analysis was performed on the input features to test the effect of dimension reduction. For the coplanar volumetric-modulated arc therapy plans, separate models were trained for voxels within the same axial slice as planning target volume voxels and voxels outside the primary beam. The effect of training separate models for each organ at risk compared to all voxels collectively was also tested. The mean squared error was calculated to evaluate the voxel dose prediction accuracy. Statistical voxel dose learning using separate models for each organ at risk had the lowest root mean squared error for all sites and modalities: 3.91 Gy (head and neck 4π), 3.21 Gy (head and neck volumetric-modulated arc therapy), 2.49 Gy (lung 4π), and 2.35 Gy (prostate volumetric-modulated arc therapy). Compared to using the original features, principal component analysis reduced the 4π prediction error for head and neck spectral regression (-43.9%) and support vector regression (-42.8%) and lung support vector regression (-24.4%) predictions. Principal component analysis was more effective in using all/most of the possible principal components. Separate organ at risk models were more accurate than training on all organ at risk voxels in all cases. Compared with more sophisticated parametric machine learning methods with dimension reduction, statistical voxel dose learning is more robust to patient variability and provides the most accurate dose prediction method.},
doi = {10.1177/1533033818811150},
journal = {Technology in Cancer Research & Treatment},
number = ,
volume = 17,
place = {United States},
year = {2018},
month = {9}
}

Journal Article:
Free Publicly Available Full Text
Publisher's Version of Record
DOI: https://doi.org/10.1177/1533033818811150

Citation Metrics:
Cited by: 2 works
Citation information provided by
Web of Science

Save / Share:

Works referenced in this record:

Feasibility of prostate robotic radiation therapy on conventional C-arm linacs
journal, July 2014


Experience-Based Quality Control of Clinical Intensity-Modulated Radiotherapy Planning
journal, October 2011

  • Moore, Kevin L.; Brame, R. Scott; Low, Daniel A.
  • International Journal of Radiation Oncology*Biology*Physics, Vol. 81, Issue 2
  • DOI: 10.1016/j.ijrobp.2010.11.030

Voxel-based dose prediction with multi-patient atlas selection for automated radiotherapy treatment planning
journal, December 2016


Predicting dose-volume histograms for organs-at-risk in IMRT planning: Predicting DVHs for OARs in IMRT planning
journal, November 2012

  • Appenzoller, Lindsey M.; Michalski, Jeff M.; Thorstad, Wade L.
  • Medical Physics, Vol. 39, Issue 12
  • DOI: 10.1118/1.4761864

Feasibility of extreme dose escalation for glioblastoma multiforme using 4π radiotherapy
journal, November 2014


Patient geometry-driven information retrieval for IMRT treatment plan quality control: Geometry-driven information retrieval for IMRT plan quality control
journal, November 2009

  • Wu, Binbin; Ricchetti, Francesco; Sanguineti, Giuseppe
  • Medical Physics, Vol. 36, Issue 12
  • DOI: 10.1118/1.3253464

Speed up kernel discriminant analysis
journal, May 2010


Viability of Noncoplanar VMAT for liver SBRT compared with coplanar VMAT and beam orientation optimized 4π IMRT
journal, January 2016


4π Noncoplanar Stereotactic Body Radiation Therapy for Head-and-Neck Cancer: Potential to Improve Tumor Control and Late Toxicity
journal, February 2015

  • Rwigema, Jean-Claude M.; Nguyen, Dan; Heron, Dwight E.
  • International Journal of Radiation Oncology*Biology*Physics, Vol. 91, Issue 2
  • DOI: 10.1016/j.ijrobp.2014.09.043

Predicting liver SBRT eligibility and plan quality for VMAT and 4π plans
journal, April 2017


Treatment planning comparison of IMPT, VMAT and 4π radiotherapy for prostate cases
journal, January 2017


Stereotactic Body Radiation Therapy in Centrally and Superiorly Located Stage I or Isolated Recurrent Non–Small-Cell Lung Cancer
journal, November 2008

  • Chang, Joe Y.; Balter, Peter A.; Dong, Lei
  • International Journal of Radiation Oncology*Biology*Physics, Vol. 72, Issue 4
  • DOI: 10.1016/j.ijrobp.2008.08.001

Stereotactic Ablative Radiation Therapy for Centrally Located Early Stage or Isolated Parenchymal Recurrences of Non-Small Cell Lung Cancer: How to Fly in a “No Fly Zone”
journal, April 2014

  • Chang, Joe Y.; Li, Qiao-Qiao; Xu, Qing-Yong
  • International Journal of Radiation Oncology*Biology*Physics, Vol. 88, Issue 5
  • DOI: 10.1016/j.ijrobp.2014.01.022

4π Non-Coplanar Liver SBRT: A Novel Delivery Technique
journal, April 2013

  • Dong, Peng; Lee, Percy; Ruan, Dan
  • International Journal of Radiation Oncology*Biology*Physics, Vol. 85, Issue 5
  • DOI: 10.1016/j.ijrobp.2012.09.028

Fully automated, comprehensive knowledge-based planning for stereotactic radiosurgery: Preclinical validation through blinded physician review
journal, November 2017

  • Ziemer, Benjamin P.; Shiraishi, Satomi; Hattangadi-Gluth, Jona A.
  • Practical Radiation Oncology, Vol. 7, Issue 6
  • DOI: 10.1016/j.prro.2017.04.011

4π Noncoplanar Stereotactic Body Radiation Therapy for Centrally Located or Larger Lung Tumors
journal, July 2013

  • Dong, Peng; Lee, Percy; Ruan, Dan
  • International Journal of Radiation Oncology*Biology*Physics, Vol. 86, Issue 3
  • DOI: 10.1016/j.ijrobp.2013.02.002

Variation in external beam treatment plan quality: An inter-institutional study of planners and planning systems
journal, October 2012


    Works referencing / citing this record:

    Artificial Intelligence in Radiotherapy Treatment Planning: Present and Future
    journal, January 2019

    • Wang, Chunhao; Zhu, Xiaofeng; Hong, Julian C.
    • Technology in Cancer Research & Treatment, Vol. 18
    • DOI: 10.1177/1533033819873922

    Radiotherapy Treatment Planning in the Age of AI: Are We Ready Yet?
    journal, January 2019

    • Zheng, Dandan; Hong, Julian C.; Wang, Chunhao
    • Technology in Cancer Research & Treatment, Vol. 18
    • DOI: 10.1177/1533033819894577

    Balancing accuracy and interpretability of machine learning approaches for radiation treatment outcomes modeling
    journal, July 2019